Biology subjects

Strauch, E.-M.

Publications and source records attributed to Strauch, E.-M..

9 recordsLinked to original sources

bioRxiv2025

ConforFold Recovers Alternative Protein Conformations Beyond MSA Subsampling

Conformational changes underlie many aspects of protein function, yet current structure prediction tools remain limited in their ability to systematically sample structural ensembles. Here, we present ConforPSSP and ConforFold, a combined framework that integrates secondary-structure sampling into deep learning-based prediction to recover multiple protein conformational states. ConforPSSP employs a transformer model trained on multi-residue fragments to generate diverse 8-state protein secondary structure predictions (PSSPs), which are then used to condition a retrained OpenFold model (ConforFold). ConforFold achieved state-of-the-art performance in conformer recovery. On our test dataset of protein samples with two alternative conformations, it correctly identified both conformers in 84% of cases at TM-scores[≥]0.8, outperforming AlphaFlow (75.4%), which uses diffusion-based sampling, and Cfold, which relies on MSA clustering. Combining ConforFold with AlphaFlow further improved recovery rates while retaining the complementary strengths of both approaches. These results establish ConforFold as a broadly applicable framework for modeling structural ensembles. By explicitly integrating secondary structure it recovers conformations inaccessible to MSA-based subsampling or diffusion models, offering a new avenue for investigating conformational heterogeneity, mechanistic transitions, and the structural basis of protein function.

biochemistry↗

bioRxiv2025

CryoPhold: CryoEM meets AlphaFold and molecular simulation to reveal protein dynamics

Here we are introducing CryoPhold, a modular workflow that unifies AlphaFold-based ensemble generation, Bayesian reweighting against experimental cryo-EM maps, molecular simulation, and machine learning to quantify conformational populations and identify structural fingerprints that govern protein functions. Proteins are inherently dynamic, interconverting among conformational states that govern their function. Perturbations such as mutations, ligand binding, and pH changes modulate these dynamics and are implicated in many diseases. While cryogenic electron microscopy (cryo-EM) has transformed structure determination, it typically yields an averaged density map representing a static snapshot. A central challenge remains capturing the thermodynamics underlying protein motions and corresponding structural fingerprints that modulate function. CryoPhold enables Bayesian reweighting of AlphaFold-generated structural ensembles against experimental cryo-EM maps, generating posterior structural ensembles that are consistent with experimental data while preserving conformational heterogeneity. Molecular simulations seeded from the posterior ensemble capture time-dependent dynamics, while machine learning models trained on featurized molecular simulation data identify structural fingerprints ("hotspots") that modulate protein dynamics. Finally, Markov state models trained on featurized molecular simulation data quantify metastable state populations and free-energy landscapes. By integrating a generative AI-based protein structure prediction model, experimental cryo-EM density, physics-based sampling, and machine learning, CryoPhold enables dynamics paradigm to capture biomolecular motion. The workflow enables prediction of equilibrium populations and structural fingerprints governing conformational dynamics in human transporter protein, GlyT1. It further captures structural changes and population shifts associated with oncogenic BRAF mutants, key driving factors behind melanoma progression.

biophysics↗

bioRxiv2024

Systematic computer-aided disulfide design as a general strategy to stabilize prefusion class I fusion proteins

Numerous enveloped viruses, such as coronaviruses, influenza, and respiratory syncytial virus (RSV), utilize class I fusion proteins for cell entry. During this process, the proteins transition from a prefusion to a postfusion state, undergoing substantial and irreversible conformational changes. The prefusion conformation has repeatedly shown significant potential in vaccine development. However, the instability of this state poses challenges for its practical application in vaccines. While non-native disulfides have been effective in maintaining the prefusion structure, identifying stabilizing disulfide bonds remains an intricated task. Here, we present a general computational approach to systematically identify prefusion-stabilizing disulfides. Our method assesses the geometric constraints of disulfide bonds and introduces a ranking system to estimate their potential in stabilizing the prefusion conformation. We found, for the RSV F protein, that disulfides restricting the initial stages of the conformational switch can offer higher stability to the prefusion state than those preventing unfolding at a later stage. The implementation of our algorithm on the RSV F protein led to the discovery of prefusion-stabilizing disulfides, providing evidence that supports our hypothesis. Furthermore, the evaluation of our top design as a vaccine candidate in a cotton rat model demonstrated robust protection against RSV infection, highlighting the potential of our approach for vaccine development. Significance statementClass I fusion proteins, pivotal for cellular entry, constitute primary targets for neutralizing antibodies across a wide range of highly pathogenic viruses. Notably, the prefusion conformation of these proteins has demonstrated to be a robust immunogen, presenting a promising avenue for vaccine development. Despite of this, the inherent structural instability of this conformation poses a substantial hurdle for vaccine formulations. To overcome this limitation and to optimize these proteins as vaccines candidates, we developed an automated computational approach to enhance the structural stability of the prefusion conformation without compromising its immunogenic properties. Given the clinical significance of class I fusion proteins, our computational tool offers a mechanism to accelerate the development of preventive measures against a multitude of viral infections.

bioengineering↗

bioRxiv2022

One-sided design of protein-protein interaction motifs using deep learning

Protein-protein interactions are part of most processes in life and thereby the ability to generate new ones to either control, detect or inhibit them has universal applications. However, to develop a new binding protein to bind to a specific site at atomic detail without any additional input is a challenging problem. After DeepMind entered the protein folding field, we have seen rapid advances in protein structure predictions thanks to the implementation of machine learning algorithms. Neural networks are part of machine learning and they can learn the regularities from their input data. Here, we took advantage of their capabilities by training multiple neural networks on co-crystal structures of natural protein complexes. Inspired by image caption algorithms, we developed an extensive set of NN-based models, referred to as iNNterfaceDesign. It predicts the positioning and the secondary structure for the new binding motifs and then designs the backbone atoms followed by amino acid sequence design. Our methods are capable of recapitulating native interactions, including antibody-antigen interactions, while they also capable to produce more diverse solutions to binding at the same sites. As it was trained on natural complexes, it learned their features and can therefore also highlight preferential binding sites, as found in natural protein-protein interactions. Our method is generally applicable, and we believe that this is the first deep learning model for one-sided design of protein-protein interactions. Abstract figure O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=120 SRC="FIGDIR/small/486144v1_ufig1.gif" ALT="Figure 1"> View larger version (37K): org.highwire.dtl.DTLVardef@6d4918org.highwire.dtl.DTLVardef@d9cd07org.highwire.dtl.DTLVardef@123b52aorg.highwire.dtl.DTLVardef@14ae130_HPS_FORMAT_FIGEXP M_FIG C_FIG

biophysics↗

bioRxiv2022

Deep learning of Protein Sequence Design of Protein-protein Interactions

MotivationAs more data of experimentally determined protein structures is becoming available, data-driven models to describe protein sequence-structure relationship become more feasible. Within this space, the amino acid sequence design of protein-protein interactions has still been a rather challenging sub-problem with very low success rates - yet it is central for the most biological processes. ResultsWe developed an attention-based deep learning model inspired by algorithms used for image-caption assignments for sequence design of peptides or protein fragments. These interaction fragments are derived from and represent core parts of protein-protein interfaces. Our trained model allows the one-sided design of a given protein fragment which can be applicable for the redesign of protein-interfaces or the de novo design of new interactions fragments. Here we demonstrate its potential by recapitulating naturally occurring protein-protein interactions including antibody-antigen complexes. The designed interfaces capture essential native interactions with high prediction accuracy and have native-like binding affinities. It further does not need precise backbone location, making it an attractive tool for working with de novo design of protein-protein interactions. AvailabilityThe source code of the method is available at https://github.com/strauchlab/iNNterfaceDesign Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗

bioRxiv2021

Robust de novo design of protein binding proteins from target structural information alone

The design of proteins that bind to a specific site on the surface of a target protein using no information other than the three-dimensional structure of the target remains an outstanding challenge. We describe a general solution to this problem which starts with a broad exploration of the very large space of possible binding modes and interactions, and then intensifies the search in the most promising regions. We demonstrate its very broad applicability by de novo design of binding proteins to 12 diverse protein targets with very different shapes and surface properties. Biophysical characterization shows that the binders, which are all smaller than 65 amino acids, are hyperstable and bind their targets with nanomolar to picomolar affinities. We succeeded in solving crystal structures of four of the binder-target complexes, and all four are very close to the corresponding computational design models. Experimental data on nearly half a million computational designs and hundreds of thousands of point mutants provide detailed feedback on the strengths and limitations of the method and of our current understanding of protein-protein interactions, and should guide improvement of both. Our approach now enables targeted design of binders to sites of interest on a wide variety of proteins for therapeutic and diagnostic applications.

synthetic biology↗

bioRxiv2021

Large-scale design and refinement of stable proteins using sequence-only models

Engineered proteins generally must possess a stable structure in order to achieve their designed function. Stable designs, however, are astronomically rare within the space of all possible amino acid sequences. As a consequence, many designs must be tested computationally and experimentally in order to find stable ones, which is expensive in terms of time and resources. Here we report a neural network model that predicts protein stability based only on sequences of amino acids, and demonstrate its performance by evaluating the stability of almost 200,000 novel proteins. These include a wide range of sequence perturbations, providing a baseline for future work in the field. We also report a second neural network model that is able to generate novel stable proteins. Finally, we show that the predictive model can be used to substantially increase the stability of both expert-designed and model-generated proteins.

biophysics↗

bioRxiv2021

Sampling of Structure and Sequence Space of Small Protein Folds

Nature only samples a small fraction in sequence space, yet many more amino acid combinations can fold into stable proteins. Furthermore, small structural variations in a single fold, which may only be a few amino acids different from the next homolog, define their molecular function. Hence, to design proteins with novel molecular functionalities, such as molecular recognition, methods to control and sample shape diversity are necessary. To explore this space, we developed and experimentally validated a computational platform that can design a wide variety of small protein folds while sampling high shape diversity. We designed and evaluated about 30,000 de novo protein designs of 7 different folds. Among these designs, about 6,200 stable proteins were identified, with predicted structures having first-of-its-kind minimalized thioredoxin. Obtained data revealed more protein folding rules, such as helix connecting loops, which were in nature. Beyond providing a resource database for protein engineering, our data presents a large training data set for machine learning. We developed a high-accuracy classifier to predict the stability of our designed proteins. The methods and the wide range of new protein shapes provide a basis for the design of new protein function without compromising stability.

bioengineering↗

bioRxiv2019

Skp1 Dimerization Conceals its F-box Protein Binding Site

Skp1 is an adapter that links F-box proteins to cullin-1 in the Skp1/cullin-1/F-box (SCF) protein family of E3 ubiquitin ligases that targets specific proteins for polyubiquitination and subsequent protein degradation. Skp1 from the amoebozoan Dictyostelium forms a stable homodimer in vitro with a Kd of 2.5 {micro}M as determined by sedimentation velocity studies, yet is monomeric in crystal complexes with F-box proteins. To investigate the molecular basis for the difference, we determined the solution NMR structure of a doubly truncated Skp1 homodimer (Skp1{Delta}{Delta}). The solution structure of Skp1{Delta}{Delta} dimer reveals a 2-fold symmetry with an interface that buries [~]750 [A]2 of predominantly hydrophobic surface. The dimer interface overlaps with subsite-1 of the F-box interaction area, explaining why only the Skp1 monomer binds F-box proteins (FBPs). To confirm the model, Rosetta was used to predict amino acid substitutions that might disrupt the dimer interface, and the F97E substitution was chosen to potentially minimize interference with F-box interactions. A nearly full-length version of Skp1 with this substitution (Skp1{Delta}F97E) behaved as a stable monomer at concentrations up to 500 {micro}M and actively bound a model FBP, mammalian Fbs1, which suggests that the dimeric state is not required for Skp1 to carry out a basic biochemical function. Finally, Skp1{Delta}F97E is expected to serve as a monomer model for high-resolution NMR studies previously hindered by dimerization.

biochemistry↗